JP6510183B2 - Method of calibrating a multiport amplifier, multiport amplifier enabling implementation of such a method, and satellite comprising such an amplifier - Google Patents

Description

  The present invention relates to a method of calibrating a multiport amplifier, which makes it possible to maximize the isolation between its outputs. The invention also relates to a multiport amplifier comprising means for implementing such a method, as well as to a satellite, in particular for telecommunications, in which the payload comprises such an amplifier.

  The invention applies mainly to the field of telecommunications, in particular space communications.

A multiport amplifier (MPA) is a system that is mainly used in the payload of a communication satellite to amplify multiple radio frequency or microwave signals, for example to implement a wide band transponder. The operating principle of the MPA is to amplify all the signals simultaneously using several amplifiers. All amplifiers, or at least some of the amplifiers, contribute to the amplification of all signals, which allows flexible allocation of power and bandwidth while ensuring optimal use of the amplifiers. This operating principle known per se is shown in FIG. FIG. 1 shows a multiport amplifier having four input ports PE1 to PE4 to which four radio frequency signals s1 to s4 are input and four output ports PS1 to PS4 from which amplified signals S1 to S4 are output. It is schematically represented. This input port is coupled to the input matrix i _i 1 to _{i i} 4 of the distribution matrix or input matrix IBM, which shows the exact same number of outputs o _i 1 to o _i 4. In this matrix, the signal si (for example, s1 represented by a vertical vector) present at the input (in this case, i _i 1) is different from each other in phase shift (indicated by a change in vector direction) It is configured to be split across all its outputs. Thus, for example, the signal s1 has no phase shift at the output o _i 1, a phase shift of 90 ° at the outputs o _i 2 and o _i 3, and a phase shift of 180 ° at the output o _i 4. The same applies to the signals s2 to s4 (not shown), even for various phase shifts. This mode of operation corresponds to the "butler" matrix known per se.

Signals arising from the output _{o i 1~o} i 4 is thus a linear combination of the input signal s1 to s4, amplified by the same power amplifier PA1 to PA4, these amplifiers, for example, traveling wave tube type (TWT It may be of the TWTA) meaning semiconductor or semiconductor type. The amplified signal is applied to the inputs i _o 1 to i _o 4 of the combination matrix or output matrix OBM, which performs operations similar to that of the distribution matrix, in the case of FIG. It is a matrix.

If the phase shift provided by the two matrices is timely selected, the vector representative of the amplified signal s1 will be zero on the outputs o _o 2 _o 4 and combined only on the output o _o 1 Can be understood in FIG. Similarly, the vector representing the amplified signal s2 is zero at the outputs o _o 1 o _o 3 and o _o 4 and is constructively coupled to only the output o _o 2 and so on. Thus, the amplified signal S _i originating from the output port Pi corresponds to the amplified input signal si and is not affected in any way by the other input signals s j of j ≠ i (i, j = 1 to 4) .

  The operation of this scheme assumes an ideal input / output matrix, and a power amplifier with exactly the same characteristics. In practice this is not the case, and the phase shift provided by the matrix is different from the nominal phase shift, such as the power of the signal at the input being split unevenly between the outputs, each amplifier exhibiting a different gain and phase, etc. There is. Furthermore, the characteristics of various elements may drift over time. Because of these discrepancies with respect to the ideal situation, the isolation between the various outputs of the multiport amplifier is not perfect, which means that the interference term corresponding to the input signal sj of j ≠ i is not It shows that it appears above.

  FIG. 2A shows the spectrum of the output signal in the case of a perfectly balanced MPA (each model of the Butler matrix is complete), whose power spectral density is expressed in dBm and the frequency (f ) Is expressed in GHz. Note that each output signal Si is an amplified version of the corresponding input signal si and is about 50 MHz wide and indicates different bands of center frequency (carrier frequency) so that the bands of the various signals do not overlap. May be In this case, it is called "disjoint frequency plan". It can be seen that the isolation is virtually complete (interference level less than -43 dB) as the interference between the signals is not visible to the naked eye. Isolation is defined as the difference (in decibels) of the integrated power in the band of the reference signal (in this case 50 MHz).

In FIG. 2B, amplitude errors and phase errors introduce random imbalances defined by a Gaussian distribution with σ _A (standard deviation of amplitude errors) = 0.8 dB and σ _φ (standard deviation of phase errors) = 3 ° The resulting spectrum is shown, and each model of the Butler matrix is "true" (obtained from measurements). The interference between the signals is clearly visible, and calculations show that the interference level is on the order of -22 dB.

  The graph in FIG. 2C shows that the imbalance in phase (.DELTA..PHI. [Degree [.degree.]] And amplitude (.DELTA.A [dB]) is an isolation IS ([dB]) between the two outputs of the multiport amplifier. It has been shown how it affects.

  In order to improve such imbalances, the amplifier of FIG. 1 comprises weighting elements EP1 to EP4, each consisting of an adjustable attenuator (rarely a preamplifier) and an adjustable phase shifter. These attenuators and phase shifters are cascaded, interlocked with their respective amplifiers, and generally connected upstream of the phase shifters. In some cases, one of the weighting factors may be omitted, as only the relative attenuation and phase shift between the outputs is relevant.

  By timely adjusting the complex weights provided by these weighting factors, it is possible to recover almost ideal, in any case greater than 20 dB of isolation between the outputs. The calibration task in adjusting these complex weights can be performed manually by the technician or preferably automatically.

  The document describes an automated method of calibrating a multiport amplifier, which maximizes isolation between outputs using measurement or test signals injected at the inputs The purpose is that. The main disadvantage of this method is that the injected signal is amplified and degrades the signal to interference ratio C / I in the telecommunication system.

  (Non-Patent Document 1) discloses an automated method of calibrating a multiport amplifier which does not require injection of a measurement signal. This method is based on the study of the probability density function of the output signal to identify and minimize interference between the outputs. This is possible only if the type of modulation is known exactly, thus limiting the flexibility of the solution.

International Publication No. 2008/135753

A paper by Mario Caron and Xinping Huang "Estimation and compensation of amplifier gain and phase mismatches in a multiple port amplifier subsystem", ESA Workshop on Advanced Flexible Telecom Payloads, November 18-20, Noordwijk (Netherlands)

  The present invention aims to provide an automated method of calibrating a multiport amplifier aiming at maximizing the isolation between the outputs, which eliminates or reduces the drawbacks of the prior art. More specifically, the present invention aims at avoiding the injection of measurement or calibration signals that are to be superimposed on the traffic signal, and any one related to the type of modulation used. It is intended not to introduce constraints.

The object of the present invention which makes it possible to achieve this goal is by means of a plurality of input ports, a plurality of power amplifiers coupled to said input ports by a matrix for distributing signals, and a matrix for recombining signals. A multiport amplifier comprising a plurality of output ports coupled to the power amplifier, wherein the matrix for distributing and recombining signals divides the signal present at the input port between the plurality of power amplifiers Configured to recombine the amplified signal at the level of the output port, and an element for weighting the signal calibrates the multiport amplifier associated with each of the power amplifiers In
a) determining an average normalized cross-correlation between an output signal present at the output port considered as a reference and signals present at the other output ports; b) repeatedly adjusting the weights of the weighting factors And B. minimizing the average normalized cross-correlation or making it less than a predetermined threshold.

According to a specific embodiment of such a method:
The average normalization cross correlation can be normalized to the average autocorrelation of the signal present at the output port considered as a reference.
The output port considered as reference can be the output port that represents the signal with the highest power level.
Said weights may be complex weights, each comprising an absolute value and a phase, said step b) comprising
b1) repeatedly adjusting the phase of the weight; and b2) repeatedly adjusting the absolute value of the weight.

More specifically, said sub-steps b1) and b2) can be carried out sequentially in any order, said sub-step b1) performing the phase of the weights of the weighting elements associated with the different power amplifiers The continuous adjustment may be included, and the sub-step b2) may include continuous adjustment of the absolute value of the weight of the weighting element associated with various power amplifiers.
Said step b) performs the following operations:
i. Determining the mean normalized cross-correlation and ii. Adding a predetermined increment to the absolute value and / or phase of the weight to be adjusted;
iii. Re-determining the average normalized cross-correlation and, if the value increases, changing the sign of the increment;
iv. Adding the predetermined increment to the phase and / or absolute value of the weight to be adjusted, together with the code determined in operation iii;
v. If the mean normalized cross-correlation is re-determined and its value decreases while maintaining the state larger than a predetermined threshold value, operation iv. And v. And b.
The average normalized correlation may be determined within a frequency band substantially coinciding with the frequency band of the signal present at the reference port.
-The signal present at the output port can indicate a non-overlapping frequency plan.
Preferably, no measurement signal is injected into the amplifier to perform a calibration.

  Another object of the present invention is to couple a plurality of input ports, a plurality of power amplifiers coupled to the input ports by a matrix for distributing signals, and a matrix for recombining signals. A plurality of output ports, the matrix for distributing and recombining the signals being configured to divide the signals present at the input ports between the plurality of power amplifiers, and of the output ports The multiport amplifier associated with each of the power amplifiers, configured to recombine the level amplified signal and for weighting the signals, to implement the calibration method as described above A multiport amplifier, characterized in that it also comprises a calibration module configured or programmed.

According to a specific embodiment of such a multiport amplifier
-Said calibration module, with first and second selection devices for selecting first and second output ports, and for selecting, converting to intermediate frequencies, filtering and digitizing said signals; First and second acquisition chains for acquiring signals present on the first and second output ports, the first and second selection devices, and the first and second acquisition chains; Driving, acquiring the digitized signal originating from the acquisition chain, based on the acquired digitized signal, the output signal acquired from the output port considered as the reference, and the other output port Calculate an average normalized cross correlation with the signal and iteratively adjust the weights of the weighting factors to minimize or determine the average normalized cross correlation And a processor programmed or configured to be less than a threshold of
-Said processors can be arranged in whole or in part, at remote locations.
-Each said weighting element may comprise a cascaded variable attenuator and variable phase shifter.
The amplifier is preferably capable of operating in at least one band chosen from the Ku band, the K band and the Ka band. In fact, the invention is even more useful the higher the frequency of the carrier, as well as when the power amplifier is a traveling wave tube. Thus, the Ka band is a promising field of application of the present invention.

  Yet another object of the invention is a communication satellite, wherein the payload comprises a multiport amplifier as described above.

  Other features, details and advantages of the invention will become apparent on reading the description which is given by way of example and with reference to the accompanying drawings. Each drawing is as follows.

1 is a schematic diagram of a multiport amplifier. FIG. 6 shows the spectrum of the output signal for a perfectly balanced MPA. It is a figure which shows the spectrum obtained by bringing in random imbalance. It is a graph which shows the influence which the unbalance has on the isolation between outputs. FIG. 1 is a basic diagram of a multiport amplifier according to an embodiment of the invention. FIG. 4 is a flow diagram of a procedure for complex weight adjustment of the weighting elements that makes it possible to maximize the isolation between the outputs of the amplifier of FIG. FIG. 4 is a spectrum of each output to the amplifier of FIG. 3 showing the technical results of the present invention. FIG. 4 is a spectrum of each output to the amplifier of FIG. 3 showing the technical results of the present invention. FIG. 5 is a schematic diagram showing the integration of a multiport amplifier according to an embodiment of the invention in a communication satellite.

  As shown in FIG. 3, the MPA according to an embodiment of the present invention receives as input a part of the signals S1 to S4 (extracted by the signal couplers CS1 to CS4) present on the output port, weighting elements EP1 to EP4. It differs from the MPA of FIG. 1 in that it comprises a calibration module MC, which generates a signal for driving EP4. More precisely, in the embodiment considered here, each weighting element is a variable attenuator AVi receiving a drive signal Ai and a variable phase shifter ViVi receiving a drive signal ii (i = 1 to 4). Equipped with

  The two selection devices SP1, SP2 each independently select the output port (more precisely, each selects the signal extracted by the signal coupler on the output port). The selected signals are processed by the respective acquisition chains CAS1, CAS2, which are usually image filters Hi, a mixer for converting each signal to an intermediate frequency, a Nyquist filter Hn, and an analog to digital conversion Device ADC. The digital signals thus obtained are processed by the processor DSP to generate drive signals A1 to A4 and Φ1 to 44. The processor DSP also drives the selection devices SP1, SP2 and the local oscillator frequency synthesizer SFL used for frequency conversion.

  The processor DSP is preferably a timely programmed digital processor (and especially a processor for digital signal processing), but other feasible means (for example products of dedicated digital circuits) are conceivable .

  According to one particular embodiment of the invention, calibration of the multiport amplifier is performed in the following manner.

  First, the processor DSP identifies the output port corresponding to the signal exhibiting the highest power. This selection is by scanning the output by one of the selection devices and by integrating the power of the acquired signal in the Nyquist band (or preferably in a narrow band selected by a digital filter). To be executed. This makes it possible to set the local oscillation frequency synthesizer SFL to the center frequency of each output, knowing the frequency plan. The output port thus identified (hereinafter PS1) is considered as a reference.

Thereafter, the first selection device SP1 is driven to select the reference port over a predetermined time window, and the second selection device SP2 is driven to select another output port over one and the same time window. This operation is repeated at all output ports. Three pairs of acquired and digitized signals, (S1 ₍₂₎ , S2), (S1 ₍₃₎ , S3), (S1 ₍₄₎ , S4) are thus obtained, where S1 _{( i)} identifies the signal S1 acquired simultaneously with Si (i = 2 to 4). This enables the processor DSP to calculate correlations or cross-correlations between the signals S1 _(i) and Si (i = 2 to 4), the maximum values of which are R ₁₁ , R ₁₂ , R ₁₃ , And R ₁₄ . The power of the signal S1 _(i) denoted by P _{1 (Ta1i)} is also calculated, eg
Is the power of the signal S1 ₍₂₎ extracted by the port PS1 at the time of synchronization acquisition of the signals S1 and S2,
Is the power of the signal S1 ₍₂₎ taken by the port PS1 at the time of synchronization acquisition of the signals S1 and S3,
Is the power of the signal S1 ₍₂₎ extracted by the port PS1 when acquiring the synchronization of the signals S1 and S4.

This allows the calculation of mean normalized cross correlation.

  The generalization to the N port case is near, where N is an integer greater than 1 and generally even.

  As further explained, the index (0) means that this belongs to the mean normalized cross-correlation determined before the first iteration of the process for adjusting the weighting factors.

  Preferably, the cross-correlation is calculated taking into account only the band of signals present on the reference port. Thus, the processor DSP drives the local oscillator frequency synthesizer SFL to select the carrier of the signal s1, and the filtering of this signal is reliably performed by sampling, but if this band is narrower than the Nyquist band, further filtering (Preferably, digital filtering by the processor DSP) can be envisaged.

  The underlying principle of the invention is to iteratively adjust the complex weights of the weighting elements EP1 to EP4 to minimize the average normalized cross-correlation or to ensure that it is at least smaller than a predetermined threshold. It is to do. In fact, the higher the average normalized cross-correlation, the lesser the outputs of the multiport amplifiers are from one another.

  This iterative adjustment can be performed using several optimization algorithms.

  One of those algorithms is described below using FIG. The principle of this algorithm is to determine if this will reduce the average normalized cross-correlation, if it does, then add the increment again, and in the opposite case, change its sign before adding the increment again. The purpose is to increase the weight phase of the weighting factor by a predetermined time interval (positive or negative). After adjusting the phases of the weighting elements in this manner, the amplitude is processed in the same manner. It is then possible to repeat this process. As a variant, it is possible to first adjust the amplitude or to adjust the amplitude and phase of one element before proceeding to the next step.

As shown in FIG. 4, the first operation of this algorithm is to determine the highest power output Sr, to measure its power, and to measure the average (normalized) cross-correlation C ₍₀₎ (Or, more specifically, calculating it based on the acquired signal). Then, this mean normalized cross-correlation is confirmatory test whether greater than a predetermined threshold value C _min, also since, if small, think that there is no need isolation is already a sufficient calculation Because it is also good.

Thereafter, the phase φ of the complex weight of the first weighting factor is incremented by the predetermined time interval Δφ, and the average (normalized) cross-correlation C ₍₁₎ is recalculated. It is then determined if the newly calculated cross-correlation C ₍₁₎ is greater than or less than C ₍₀₎ , and if so, this means that the phase increment has increased instead of reducing the imbalance. This means that the sign of the incremental time interval is changed to Δφ → −Δφ, and if small, this time interval remains unchanged. Then, after successive occurrences of an increment (always in the exact same time interval, whose sign is determined only once after the described first repetition), one of the following conditions is fulfilled: Be
The mean cross-correlation falls below the threshold C _{min, in} which case the isolation is considered sufficient and the process is considered to be halted.
Or alternatively, the average cross-correlation starts to increase (C _{(n + 1)} > C _(n) ), which indicates that the optimal set point has been exceeded, where Δφ is subtracted from the phase of the complex weight, Returning to the optimal setting identified, the next weighting factor is adjusted continuously (for the sake of simplicity, the flow diagram of FIG. 1 is limited to the case of a single weighting factor).

If all phases are adjusted and the average cross-correlation is above the threshold _Cmin , the amplitude is adjusted in the same manner (right hand side of the flow diagram).

  Then, if the isolation is still not sufficient, repeat again (not shown).

  Of course, it is possible not to exceed the maximum number of repetitions, whether with regard to the adjustment of the phase and / or the amplitude of each element, or with regard to the procedure as a whole.

5A-5C make it possible to show the technical results of the present invention. The figures show the spectra of the output signals of the multiport amplifier of FIG. 3, which are obtained by numerical simulation under the following conditions.
The imbalance of amplitude and phase follows a Gaussian distribution of σ A = 0.8 dB and σ φ = 3 °.
-Dynamic variation on the power of the carrier = 3 dB.
2 Acquisition of signals at ¹³ points.
-Band limited to 50 MHz (sampling at 100 MHz).
A QPSK modulated carrier with a bandwidth of 20 MHz at each input.
Carrier waves equally divided between 50 MHz and 200 MHz.
-A model of the Butler matrix resulting from measurements on the actual device.
No modeling of the AM / AM effect of the power amplifier (gain variation at the operating point with respect to power) and AM / PM effect (phase variation of the transfer function at the operating point with respect to power).

  In FIG. 5A, the spectrum before compensation is shown and it can be seen that the interference is significant and indeed the worst case isolation is -16.8 dB, which is generally unsatisfactory. After five iterations (FIG. 5B), the inter-carrier interference virtually disappears and the worst case isolation is equal to -31.9 dB. In fact, it is possible to already make the minimum isolation a value of -22.6 dB with only one iteration, which is generally considered sufficient for most applications.

  FIG. 6 very schematically shows the integration of the multiport amplifier MPA according to the invention with the payload CU of the communication satellite SAT. The payload CU is a transponder comprising a transmitter and a receiver that share one and the same antenna ANT, and the multiport amplifier constitutes the power stage of this transmitter.

Although the invention has been described with respect to particular embodiments, a number of variations are conceivable, such as:
-The multiport amplifier can be used in applications other than space communication and different spectral bands than the spectral bands (Ku, K, Ka bands) considered here, eg L, S, C, And works with X band.
-In particular, the input / output matrix may not be a Butler matrix, if it is necessary to use multiple ports which are not a power of two, and thus multiple power amplifiers.
The number of amplifiers is different from the number of signals to be amplified, and may in particular be more than that number.
Exceptionally, the calibration can be carried out exclusively by acting on the phase or the absolute value of the weight of the weighting element.
-The calibration method may be performed by using the original signal processed by the multiport amplifier in normal operation, and if the original signal is not present, it has been modulated with the test signal (or indeed it is It is possible to use unmodulated) carriers, or just noisy carriers.
-The signals processed by the multiport amplifier may for example use the exact same carrier without showing a non-overlapping frequency plan, but in this case the isolation between the outputs is better than in the case of a non-overlapping frequency plan It can be worse.
-The reference output port may not be the output port that represents the signal with the highest power, but this embodiment is preferred.
-Several different optimization algorithms may be used to adjust the weight of the weighting factor.
-The processing of the signals may be located at least partially apart, for example, the signals acquired and digitized by the acquisition chain may be stored in memory, perform calculations and weighting elements A control signal may be transmitted back to the ground station to perform the adjustment.

s1 radio frequency signal s2 radio frequency signal s3 radio frequency signal s4 radio frequency signal PE1 input port PE2 input port PE3 input port PE4 input port S1 amplified signal S2 amplified signal S3 amplified signal S3 amplified signal PS1 output Port PS2 output port PS3 output port PS4 output port i _i 1 input i _i 2 input i _i 3 input i _i 4 input o _i 1 output o _i 2 output o _i 3 output o _i 4 output PA1 power amplifier PA2 power amplifier PA3 power Amplifier PA4 Power amplifier i _o 1 input i _o 2 input i _o 3 input i _o 4 input o _o 1 output o _o 2 output o _o 3 output o _o 4 output EP 1 weighting element EP 2 weighting element EP 3 weighting element EP 4 weighting element EP 4 weighting element CS 1 signal Coupler CS2 Signal coupler CS3 Signal coupler CS4 Signal coupler SP1 Selection device SP2 Selection device CAS1 Acquisition chain CAS2 Acquisition chain A1 Drive signal A2 drive signal A3 drive signal A4 drive signal 11 drive signal 22 drive signal 33 drive signal 44 drive signal

Claims (14)

A plurality of input ports (PE1 to PE4), a plurality of power amplifiers (PA1 to PA4) coupled to the input ports by a matrix (IBM) for distributing the signal, and a matrix (OBM for recombining the signal A plurality of output ports (PS1 to PS4) coupled to the power amplifier by the matrix), the matrix for distributing and recombining a signal, the input between the plurality of power amplifiers Said signal distributed at the level of said output port and configured to recombine said signal amplified at the level of said output port and distributed by a matrix (IBM) for distributing the signal An element for weighting at least one of the amplitude and the phase of the signal (EP1 to EP4) is the power amplification A method of calibrating a multi-port amplifier associated with,
a) selecting a reference output port from the output port;
b ) determining an average normalized cross-correlation between the output signal present at the reference output port and the signals present at the other output ports;
c ) iteratively adjusting the weights of the weighting elements to minimize the average normalized cross-correlation or to bring it below a predetermined threshold.   The method according to claim 1, wherein the output port considered as a reference is an output port indicating a signal with the highest power level. Said weights are complex weights, each comprising an absolute value and a phase, and said step c ) comprises
c 1) repeatedly adjusting the phase of the weight repeatedly;
c 2) repeatedly adjusting the absolute value of the weight in the substep of c . Said sub-steps c 1) and c 2) are performed sequentially in any order, said sub-step c 1) continuously adjusting said phases of said weights of said weighting elements associated with said various power amplifiers The method according to claim 3, wherein the substep c2 ) comprises a continuous adjustment of the absolute value of the weight of the weighting element associated with the various power amplifiers. The above step c ) performs the following operation:
i. An operation of determining said average normalized cross correlation;
ii. Adding a predetermined increment to the absolute value and / or phase of the weight to be adjusted;
iii. Re-determining the average normalized cross-correlation and, if the value increases, changing the sign of the increment;
iv. Adding the predetermined increment to the phase and / or the absolute value of the weight to be adjusted, along with the sign determined in operation iii;
v. If the mean normalized cross-correlation is re-determined and its value decreases while maintaining the state larger than a predetermined threshold value, operation iv. And v. The method according to any one of claims 1 to 4, further comprising:   The method according to any of the preceding claims, wherein the average normalized cross correlation is determined within a frequency band substantially coinciding with the frequency band of the signal present at the reference port.   7. A method according to any one of the preceding claims, wherein the signal present at the output port indicates a non-overlapping frequency plan.   The method according to any one of the preceding claims, wherein no measurement signal is injected into the amplifier to perform the calibration. A plurality of input ports (PE1 to PE4), a plurality of power amplifiers (PA1 to PA4) coupled to the input ports by a matrix (IBM) for distributing the signal, and a matrix (OBM for recombining the signal And a plurality of output ports (PS1 to PS4) coupled to the power amplifier by: the matrix for distributing and recombining the signals present at the input port between the plurality of power amplifiers. It is configured to divide and is configured to recombine the amplified signal at the level of the output port, and an element for weighting the signal (EP1 to EP4) is associated with each of the power amplifiers In the port amplifier
A multiport amplifier, characterized in that it also comprises a calibration module (MC) configured or programmed to carry out the calibration method according to any one of the preceding claims. The calibration module
-First (SP1) and second (SP2) selection devices for selecting the first and second output ports;
-A first (CAS1) and a second (CAS1) for acquiring the signals present on the first and second output ports for selecting, converting to intermediate frequencies, filtering and digitizing the signals With acquisition chain of CAS 2),
-Driving the first and second selection devices and the first and second acquisition chains, acquiring digitized signals originating from the acquisition chains, based on the acquired digitized signals Calculating an average normalized cross-correlation between an output signal obtained from an output port considered as a reference and signals obtained from other output ports, repeatedly adjusting the weight of the weighting factor, and calculating the average normal A multiport amplifier according to claim 9, comprising a processor (DSP) programmed or configured to minimize the cross-correlation or to bring it below a predetermined threshold.   11. A multiport amplifier according to claim 10, wherein the processor is located in whole or in part at a remote location.   A multiport amplifier according to any one of claims 9 to 11, wherein each of the weighting elements comprises a variable attenuator (AV1 to AV4) and a variable phase shifter (VV1 to VV4) connected in cascade.   The multiport amplifier according to any one of claims 9 to 12, which operates in at least one band selected from a Ku band, a K band, and a Ka band.   A satellite (SAT) for telecommunications, wherein the payload comprises a multiport amplifier according to any one of claims 9 to 13.